The finite-step realizability of the joint spectral radius of a pair of d×d matrices one of which being rank-one

We study the finite-step realizability of the joint/generalized spectral radius of a pair of real d × d matrices {S1, S2}, one of which has rank 1, where 2 ≤ d < +∞. Let ρ(A) denote the spectral radius of a square matrix A. Then we prove that there always exists a finite-length word (i1, . . . , i ∗ l) ∈ {1, 2}, for some finite l ≥ 1, such that l √ ρ(Si1 · · ·Si∗l ) = sup n≥1 { max (i1,...,in)∈{1,2} n √ ρ(Si1 · · ·Sin ) } ; that is to say, there holds the spectral finiteness property for {S1, S2}. This implies that stability is algorithmically decidable for {S1, S2}.